{ "name": "Multiplet", "updated": "2026-10-18", "description": "A complete AI fleet run for the project you already work on with a coding agent.", "availability": "Early-access signups are open. Public self-service Console, API and checkout are coming soon.", "audience": "People using coding agents for substantial engineering, research and analysis projects.", "intended_workflow": [ "Give your existing agent the live Multiplet documentation", "Choose a first fleet or monthly plan when access opens", "Let your agent prepare the task, files, tools and configuration", "Approve access, the run estimate and start time", "Let the fleet investigate, build and share discoveries", "Retrieve code, tests, experiments and findings into the original project" ], "capabilities": [ "Working Linux environments with code execution", "Leading open-weight models served with the agent harness", "Direct messages and shared project files", "Agent-organized roles and parallel approaches", "Different environment sizes for different work", "Project files and available activity returned to your existing agent" ], "positioning": { "offer": "Give your coding agent a complete fleet.", "economics": "First fleet: 256 agents for four hours at $1,000. Monthly: twenty standard runs for $10,000. The service brings model inference, the harness, computers and fleet operation together.", "workflow": "Use the coding-agent session that already knows the project to prepare work and inspect what returns." }, "pricing": { "schema_version": 3, "updated": "2026-10-17", "currency": "USD", "credit_price": 0.5, "credits_per_agent_hour": 1, "purchase_unit": { "name": "Fleet run", "agents": 256, "hours": 4, "credits": 1024, "configuration": "Standard model and environment" }, "first_fleet": { "price": 1000, "agents": 256, "hours": 4, "configuration": "Standard model and environment", "included_runs": 1, "billing_period": "one_time", "subscription_required": false, "automatic_renewal": false, "service_failure": "Replacement run or refund if a Multiplet technical failure prevents delivery of the purchased run; no guaranteed research outcome", "upgrade_credit": { "amount": 1000, "within_days": 30, "applies_to": "First monthly invoice", "first_month_invoice": 9000, "additional_credits": 0 } }, "monthly": { "price": 10000, "included_runs": 20, "included_credits": 20480, "billing_period": "month", "allowance": "20 standard fleet runs per month", "top_up_price_per_credit": 0.5 }, "enterprise": { "name": "Enterprise", "pricing": "Contact us", "contact_url": "/contact/", "scope": "Custom run volume and configuration by agreement" }, "status": "Launch pricing", "billing_live": false, "purchase_flow": "Self-service Console; public checkout forthcoming", "required_sales_call": false, "first_fleet_is_wallet_credit": false, "offer_pages": [ "/first-fleet/", "/pricing/" ], "estimate_is_exact_meter": false, "planning_basis": "Approximately one credit per agent-hour for a standard configuration", "pays_for": [ "Model inference", "Working environments", "Fleet coordination and operation" ], "spending": "Additional purchases require approval; no automatic top-ups", "metering_status": "The first fleet is a fixed standard package. Credits account for monthly allowance and approved extra usage. Different model/environment configurations can change consumption; final metering and settlement terms will be shown before checkout opens", "terms_status": "Exact metering, credit validity, cancellation, retention and service-failure terms must be finalized before checkout opens", "standard_resources": { "environment": "One working computer per agent; CPU capacity is shared and allocations can be fractional.", "configured_resources": "The model, CPU reservation and memory guarantee are confirmed before purchase.", "meter_scope": "Current resource estimate for one agent per computer; final metering terms confirmed before checkout.", "resource_rate_formula": "max(1, reserved_cpu_cores, guaranteed_memory_MiB / 2048) credits per agent-hour", "examples": [ { "reserved_cpu_cores": 0.25, "guaranteed_memory_MiB": 768, "credits_per_agent_hour": 1 }, { "reserved_cpu_cores": 1, "guaranteed_memory_MiB": 2048, "credits_per_agent_hour": 1 }, { "reserved_cpu_cores": 1, "guaranteed_memory_MiB": 4096, "credits_per_agent_hour": 2 } ], "burst": "Extra vCPUs and opportunistic CPU or memory burst above the reservation do not increase the resource rate; spare capacity is not guaranteed." } }, "models": { "family": "Leading open-weight models", "options": [ { "name": "GLM-5.3", "developer": "Z.ai", "source": "https://huggingface.co/zai-org/GLM-5.3" }, { "name": "MiMo-V2.6-Pro", "developer": "Xiaomi", "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" }, { "name": "DeepSeek-V4.1-Flash", "developer": "DeepSeek", "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash" } ], "experiment_models": "Models vary by run. Each published result states its model where recorded.", "selection": "Model and configuration are confirmed before purchase. Ask us about a preferred model.", "public_self_service_available": false, "human_page": "/features/models-and-tools/" }, "live_interaction": { "feature": "Interact with a running fleet", "status": "Coming soon", "public_self_service_available": false, "interfaces": [ "Your existing coding agent", "Multiplet interface" ], "actions": [ "Follow work in progress", "Send new context", "Ask follow-up questions", "Redirect the work" ], "human_page": "/features/observability/" }, "current_customer_action": { "type": "Early access", "url": "/early-access/", "payment_taken": false, "capacity_reserved": false, "immediate_self_service_run_access": false }, "delivery": { "contents": "Project files, code, tests, experiments, findings and available run activity. Exact export contents and retention are confirmed before purchase.", "review": "Your agent can inspect files, rerun checks and help apply useful changes." }, "evidence": [ { "url": "/evidence/java-migration/result.json", "human_page": "/examples/java-migration/", "result": "Returned four-file patch completed the prepared Restlight Java 17 build with 729 original tests passing and 219 test files unchanged." }, { "url": "/evidence/compression-result.json", "human_page": "/examples/compression/", "result": "Selected Zstd implementation improved throughput by 7.81% versus the pinned reference across six public Silesia files; compressed sizes and decoding checks passed." }, { "url": "/evidence/circuit-result.json", "human_page": "/examples/circuit-design/", "result": "Selected Ibex ALU design reduced estimated Balanced area by 23.58% and delay by 35.01%; formal equivalence passed all three tested configurations." }, { "url": "/evidence/coordination/java-observations.json", "human_page": "/examples/self-organization/", "result": "Recorded repository claims and diagnosis handoffs during a 32-agent Java migration." } ], "human_pages": [ "/features/", "/how-it-works/", "/examples/", "/pricing/", "/first-fleet/", "/compare/", "/security/" ], "detail_pages": { "models": "/features/models-and-tools/", "data_handling": "/security/", "billing": "/pricing/", "result_methods": "/benchmarks/methodology/", "component_cost_model": "/compare/costs/" }, "roadmap": [ { "feature": "Interact with a running fleet", "status": "Coming soon", "public_self_service_available": false, "interfaces": [ "Your existing coding agent", "Multiplet interface" ], "actions": [ "Follow work in progress", "Send new context", "Ask follow-up questions", "Redirect the work" ], "human_page": "/features/observability/" }, { "feature": "Affordance open-source release", "status": "Planned" }, { "feature": "Adaptive intelligence", "status": "In development; not included in the current offer", "intent": "Specialize model weights during a run and return the resulting weights", "measured_benefit_established": false } ] }